Nila AI Assistant

✅ The onboarding path

💬 Chat — /nila

Talk to your assistant in the browser. Same brain as the Telegram bot, with a full conversation workspace around it.

Nila on Telegram

The same assistant lives on Telegram. To link your account:

1. In Telegram, search for @nila_xq_bot and tap Start
2. In Nila, open Settings → Connect Telegram — you'll get a one-time code
3. Send the code to the bot: /start <code> (or use the one-tap link button)

The code is valid for 15 minutes. Once linked, just message the bot — it's your Nila, with the same memory and conversations as the web app.

🎯 Full vs Direct — which mode when?

The toggle at the top right of Chat picks how much of Nila wakes up for that conversation. Think of it this way: Full is your assistant sitting at its desk with your calendar, inbox and to-do list open. Direct is the same brain on a walk with you — great conversation, but nothing to look things up in and no way to act.

The same message, two different outcomes

📁 Files — /files

Your assistant's file home — everything it can search, ground answers in, or has filed for you lives here.

From pocket to filed, in one Telegram message

Business cards become a pocket CRM

⚙ Control — /control

The cockpit. Connect services, pick models, shape the personality, and watch the system run — the guides below all end here.

Overview

One API key unlocks every AI capability. You then pick which model handles each job from the Control panel — and can swap them anytime.

Provider

Prerequisites

• An Alibaba Cloud account with Model Studio activated.
• The region you'll use — International (Singapore) or China (Beijing). Models and the endpoint URL differ per region.
• Access to either the server's .env file or the Control panel to store the key.

1 Get your Model Studio API key

1. Sign in to the Alibaba Cloud Model Studio console (International). If it's your first time, activate Model Studio when prompted.
2. Open API Keys (under your profile / "API-KEY" menu).
3. Click Create API Key, then copy the value — it starts with sk- and is shown once.
4. Note your region's base URL (OpenAI-compatible mode):

   Region	Base URL
   International (Singapore)	https://dashscope-intl.aliyuncs.com/compatible-mode/v1
   China (Beijing)	https://dashscope.aliyuncs.com/compatible-mode/v1

2 Add the key to Nila

As with the other connectors, the key lives in two places — the worker (for chat, media, and briefings) and n8n (for workflow AI calls).

Option A — Environment file (.env)

On the server, edit /home/g-os/nila/.env:

# --- AI Engine (Alibaba Cloud Model Studio, OpenAI-compatible) ---
DASHSCOPE_API_KEY=sk-your-model-studio-key
DASHSCOPE_BASE_URL=https://dashscope-intl.aliyuncs.com/compatible-mode/v1
DASHSCOPE_MODEL=qwen-plus   # default chat model (fallback)

Then restart the worker so it picks up the new variables:

cd /home/g-os/nila && docker compose up -d worker

Option B — Control panel

Open the Control panel, find AI Engine, and edit the credential. Paste the API key and the base URL for your region, then save.

3 Choose your models

In the Control panel → Choose your AI models, pick a model for each job from the dropdowns (the list is pulled live from your connected account), then Save Models. Changes apply instantly — no restart.

4 Verify the connection

1. Open the Control panel — the model section should show "N models available" and populated dropdowns. That confirms the key and base URL work.
2. Send a normal chat message (Telegram or Nila) and confirm you get an AI reply.
3. Try /imagine a sunrise over Jakarta to confirm image generation.

Troubleshooting

Overview

Once connected, you message a private Telegram bot and it routes your request through the assistant — fetching context, calling the AI, and replying right in the chat. What you can do:

Supported channels

Prerequisites

• A Telegram account (the phone app or desktop client).
• Access to either the server's .env file or the Control panel to store the bot token.
• The assistant stack running, with n8n reachable at your domain (it hosts the bot's webhook).

1 Create a Telegram bot

1. In Telegram, open a chat with @BotFather (the official bot for creating bots).
2. Send /newbot.
3. Give it a display name (e.g. Nila Assistant).
4. Give it a username ending in bot (e.g. nila_assistant_bot). It must be unique.
5. BotFather replies with a token like 123456789:AAH...xyz. Copy it.

2 Get your Telegram user ID

The bot is private — it should only answer you. Your numeric user ID is how it knows that.

1. Open a chat with @userinfobot and send any message.
2. It replies with your Id — a number like 123456789. Copy it.

3 Add the token to Nila

The same bot token is used in two places — the worker (for outbound messages like briefings and confirmations) and n8n (for receiving your messages).

Option A — Environment file (.env)

On the server, edit /home/g-os/nila/.env:

# --- Telegram Bots ---
PERSONAL_BOT_TOKEN=123456789:AAH...your-bot-token
YOUR_TELEGRAM_ID=123456789

Then restart the worker so it picks up the new variables:

cd /home/g-os/nila && docker compose up -d worker

Option B — Control panel

Open the Control panel, find Chat Interface → Telegram, and click to edit the credential. Paste the same bot token and save. This is the credential n8n uses to receive messages.

4 Activate the bot (register the webhook)

A Telegram bot delivers messages to one webhook URL. n8n registers that webhook automatically when you activate the routing workflow — no manual setWebhook call needed.

1. Open n8n at https://nila.xqui.site and sign in.
2. Open the Personal Bot Main Router workflow.
3. Confirm the Telegram Trigger node uses the credential from Step 3.
4. Toggle the workflow Active (top-right). n8n registers the webhook with Telegram on activation.

curl -s "https://api.telegram.org/bot<BOT_TOKEN>/getWebhookInfo"

5 Verify the connection

1. Open your new bot in Telegram (search its @username) and press Start.
2. Send "hi" — it should reply within a few seconds.
3. Try "what can you do?" or /imagine a sunset over Jakarta to confirm the full pipeline works.

Commands & usage

You can just talk to the bot — it detects what you want. A few explicit slash commands are also available.

Slash commands

Just say it (natural language)

Troubleshooting

Overview

Once connected, the assistant talks to your mailbox and calendar across Telegram, the Nila chat UI, and scheduled workflows (such as the morning briefing). Here is what the connection enables:

What it looks like in chat

Supported providers

Prerequisites

• A Microsoft 365 / Entra ID (Azure AD) tenant.
• Global Administrator (or Privileged Role Admin + Cloud App Admin) rights — application permissions require admin consent.
• The mailbox address the assistant should act on, e.g. galih@yourdomain.com.
• Access to either the server's .env file or the Control panel to store credentials.

1 Register an app in Azure

1. Go to the Microsoft Entra admin center (or Azure portal) and sign in as an administrator.
2. Open Identity → Applications → App registrations and click New registration.
3. Name it something recognizable, e.g. Nila Assistant.
4. Under Supported account types, choose Accounts in this organizational directory only (single tenant).
5. Leave Redirect URI empty — this is an app-only connection, no browser sign-in. Click Register.
6. On the app's Overview page, copy the Application (client) ID and the Directory (tenant) ID.

2 Grant API permissions

The connection uses Application permissions (app-only), not Delegated. This lets background workflows run without a signed-in user.

1. In your app, open API permissions → Add a permission.
2. Choose Microsoft Graph → Application permissions.
3. Add each of the permissions in the table below.
4. Back on the API permissions page, click Grant admin consent for <your tenant> and confirm. Each row should show a green Granted state.

3 Create a client secret

1. In your app, open Certificates & secrets → Client secrets → New client secret.
2. Give it a description (e.g. nila-prod) and choose an expiry (24 months is typical).
3. Click Add, then immediately copy the secret's Value (not the Secret ID).

4 Collect your four values

You should now have everything the assistant needs:

5 Add the credentials to Nila

There are two places these values live. Use whichever matches how the feature is driven.

Option A — Environment file (.env)

Powers the Python worker integrations (Telegram bot, Nila chat, briefings). On the server, edit /home/g-os/nila/.env:

# --- Microsoft 365 (Graph API) ---
MS_TENANT_ID=your-directory-tenant-id
MS_CLIENT_ID=your-application-client-id
MS_CLIENT_SECRET=your-client-secret-value
MS_USER_EMAIL=galih@yourdomain.com

Then restart the worker so it picks up the new variables:

cd /home/g-os/nila && docker compose up -d worker

Option B — Control panel

For n8n-driven workflows, open the Control panel, find Email & Calendar → Microsoft 365, and click to edit the credential. Paste the same Tenant ID, Client ID, and Client secret, then save.

6 Verify the connection

Confirm the credentials work end-to-end with two quick checks.

From the assistant (easiest)

• Message the bot: "What's on my calendar today?" — it should list today's events.
• Message the bot: "Any unread emails?" — it should return recent unread mail.

From the server (direct API)

# today's calendar events
curl -s https://nila.xqui.site/api/calendar/today

# unread email
curl -s https://nila.xqui.site/api/mail/unread

Permissions reference

Exactly which Graph permission each capability relies on:

Troubleshooting